摘要
为进一步提高传统极限学习机的泛化能力,提出了一种基于人工蜂群算法优化的极限学习机模型。该模型将人工蜂群算法的全局寻优能力和极限学习机的快速学习能力相结合,有效克服了传统极限学习机的过拟合现象。在确定水压变化比值作为故障特征参数的基础上,将优化后的极限学习机模型应用于供水管网的泄漏故障诊断实验,实验结果表明,经人工蜂群算法优化的极限学习机模型在故障诊断速度和精度方面均优于其他3种模型。
To improve the generalization performance of the traditional extreme learning machine,an extreme learning machine model optimized is presented by artificial bee colony algorithm.The global optimization ability of artificial bee colony algorithm is combined with quick learning ability of extreme learning machine in this model,and the over-fitting phenomenan in traditional extreme learning machine are overcome efficaciously.On the basis of choosing the relative change value of water pressure as characteristic parameter,the optimized extreme learning machine model is applied to leakage fault diagnosis experiment of water supply network,and experimental results show that the fault diagnosis speed and precision of optimized extreme learning machine based on artificial bee colony algorithm are better than the other three models.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第8期2887-2891,共5页
Computer Engineering and Design
基金
国家"十一五"科技支撑计划重点基金项目(2006BAJ16B08)
关键词
极限学习机
优化算法
故障诊断
供水管网
人工蜂群算法
extreme learning machine
optimization algorithm
fault diagnosis
water supply network
ABC algorithm